A Visual History of Loudness in Popular Music

January 5, 2010  |  Infographics

loudness

All Things Considered discusses why music sounds worse than it did a few decades ago. Through a practice using compressors, the quiet parts of a song are made louder and the louder parts quieter so that the song as a whole sounds louder to your ear. The purpose: to make the song stand out when you hear it on the radio.

As a result, tracks have gotten louder over the years.

A Trend Towards Loud

Christopher Clark explores the trend of loudness with this graphic [pdf]. He selected several songs from each year and summed the maximum RMS levels. Songs are color-code by genre.

What you see is an upward trend towards louder, and what some might argue, suckier music.

Had it not been for Robert Siegel's chat with Bob Ludwig, a record mastering engineer, about this trend, I'd be more skeptical about what I'm seeing. But Ludwig's explanations follow the visual, so just disregard the nonsensical pie charts on the bottom and the statistical bias that I am sure stemmed from hand-picking the songs.

Oh how I long for the the music of the 80s, when music was pure and beautiful and so much easier on the ears.

You know you like it.

[Thanks, Ben]

27 Comments

  • “Oh how I long for the the music of the 80s, when music was pure and beautiful and so much easier on the ears.”

    Yeah, right. The 80s were mostly a musical wasteland.

    • i actually wouldn’t know. i was listening to my parents’ motown favorites up through the early 90s. i get the feeling though that i missed out something amazing.

  • Eric Cherry January 5, 2010 at 1:01 pm

    I think you need to go back a little further than the 80′s for good music. My 15 year old son is now into Pink Floyd, The Who, Jethro Tull and early Genesis. Thought provoking lyrics, complex overlays, classical/symphonic underpinning, and guitar/synthesizer effects that evoke emotion and imagry.

  • Anyone who thinks there wasn’t good music in the 80′s must’ve just been listening to Top 40 radio.

  • Wait. What? No.

    The graph shows *peak* recording levels increasing, which would result in more accurate playback (due to the full resolution of being used). This has *nothing* to do with any changes to *softer* parts of the song, which aren’t shown in the graph. This is not even to mention the obvious opportunity for bias in songs selected for analysis.

    And ultimately, loudness is determine by your volume knob (or button) which still work as well as they did in 1979.

    • Higher peak recording levels don’t necessarily more accurate. It just means louder.

      And yes, you can always adjust volume with the knob, but you can only transmit a certain level over the radio waves. They don’t give more “bandwidth” for louder music, which is the point of these tricks. It’s more clear if you listen to interview with Ludwig.

      • Given a continuous waveform encoded digitally, scaling the waveform such that the peak amplitude utilizes the full digital resolution is a more accurate representation of the original analog waveform than scaling it to use a fraction of the resolution. Scaling beyond will result in clipping the signal and a distortion of the original sound pressure wave.

        Based on the area of the poster labeled “dynamic”, the author seems to be assuming that digital recording resolution was maximized in the early nineties and that any increases since then have resulted in clipping the signal. Indeed, this seems to be the primary criticism from the All Things Consisdered article (especially the discussion of the Beatles and Metallica).

      • “Given a continuous waveform encoded digitally, scaling the waveform such that the peak amplitude utilizes the full digital resolution is a more accurate representation of the original analog waveform than scaling it to use a fraction of the resolution”

        Only with regard to dynamic range, really. It’s not “more accurate”, unless the given material has a dynamic range that exceeds what is possible in a given format. (A “more accurate representation of the…waveform” is also a function of sample rate, i.e. how many times per second a waveform is sampled – 44,100 samples/second is the norm for CDs, with 16 bits dynamic range.)

        A “bit” can represent 6 dB (a fourfold increase) of dynamic range, so a 16-bit system can represent 16×6, or 96 dB of dynamic range. Popular music may have 10-15 dB of dynamic range, which can actually be represented by 3 bits – and it doesn’t really matter which 3 they are . At that point, you really are just talking about making it louder, not more accurate.

        Other forms of music have greater dynamic range, but 96 dB is a pretty wide range, and why most people find it acceptable. Step up to 24-bit audio, and you can represent 144 dB of dynamic range, which is approximately equal to the full range of human hearing.

      • Yes, only with regard to dynamic range. All other things (including recording equipment, sampling rate, etc, etc) assumed to be equal.

    • A simple experiment is enough – if you own both a cd player and an old-fashioned analogue record-player, and a cd and vinyl copy of the same album.
      Switch the source on your amplifier from “cd” to “phono” (or whatever input serves the record player) at similar points of the same track.
      The cd will sound louder but the extremes will have been surpressed – or deleted totally – to save digital disc space.
      The record’s playback volume will need turning up to match the actual sound emanating from the speakers, and you should hear more levels or tracks, such as hi-hat, cymbals, etc at one end, and more bass, kick-drum, etc at the other.
      Nowt to do with graphs and that, just personal experience.

  • Now plot sales against loudness.

    • they aren’t peak levels.. it is mentioned that they are RMS levels which aren’t peaks, but the average levels.. which indeed have become more and more limited in what is being deemed: “The Loudness Wars”.. It’s not a mystery.. it’s a fact. The thing that most people are critical of in this trend is the loss of dynamics.. Mastering engineers receive pressure to make the singles as loud as the next for radio.. and it becomes an endless cycle.

  • The 80ies were fine but Rick Assh… was one of it’s palgues. I nowadays listen to Noise & Industrial. There is a lot of dynamics in the tracks…

  • The 70′s were my favorite decade of music….Women were finally in the mix and regarded as equals…..well they were in the mix..they’re still fighting for equal status..but long live the music of Carly Simon, Carole King, Joni Mitchell, Janis Ian, Laura Nyro!

  • I actually wrote about this concept last year, often known as “The Loudness War”: http://lithe.wordpress.com/2009/08/24/the-loudness-war/

    I’m not sure how we can even begin to reverse the effect…

  • Why don’t you add a ‘re-tweet’ button to your amazing articles? I would make them so much easier to share!

  • This is why I stay away from mainstream releases coming out now. Independent labels and others that have no chance of having a single played on radio tend to produce music at the volume it should be mastered at.

Copyright © 2007-2014 FlowingData. All rights reserved. Hosted by Linode.